Background Pathological myopia is one of the leading causes of blindness globally. Lower birth weight (BW) within the normal range has been reported to increase the risk of myopia, although findings conflict. We sought to estimate the causal effect of BW on refractive error using Mendelian randomisation (MR), under the assumption of a linear relationship. Methods Genetic variants associated with BW were identified from meta-analysis of a genome-wide association study (GWAS) for self-reported BW in 162 039 UK Biobank participants and a published Early Growth Genetics (EGG) consortium GWAS (n=26 836). We performed a one-sample MR analysis in 39 658 unrelated, adult UK Biobank participants (independent of the GWAS sample) using an allele score for BW as instrumental variable. A two-sample MR sensitivity analysis and conventional ordinary least squares (OLS) regression analyses were also undertaken. Results In OLS analysis, BW showed a small, positive association with refractive error: +0.04 D per SD increase in BW (95% CI 0.02 to 0.07; p=0.002). The one-sample MR-estimated causal effect of BW on refractive error was higher, at +0.28 D per SD increase in BW (95% CI 0.05 to 0.52, p=0.02). A two-sample MR analysis provided similar causal effect estimates, with minimal evidence of directional pleiotropy. Conclusions Our study suggests lower BW within the normal range is causally associated with a more myopic refractive error. However, the impact of the causal effect was modest (range 1.00 D covering approximately 95% of the population).
